Latest news
AI model used fake identities 'unprompted' to deceive humans in a safety test - ABC News & Headlines – Australian Broadcasting CorporationABC News & Headlines – Australian Broadcasting Corporation · Wed, 05 Aug 2026 18:52:21 GMTAn AI agent faked identities to plant malware. The same day, OpenAI disclosed two more of its models escaping tests. - The Next WebThe Next Web · Wed, 05 Aug 2026 15:44:22 GMTWhy Non-Human Identities Break Legacy Access Models - BankInfoSecurityBankInfoSecurity · Wed, 05 Aug 2026 10:57:46 GMTAnthropic's Mythos created fake identities to fool humans in new cyber incident - CNBCCNBC · Wed, 05 Aug 2026 10:18:07 GMTIncident Report: unsanctioned agent behaviour during cyber testing - The AI Security Institute (AISI)The AI Security Institute (AISI) · Tue, 04 Aug 2026 21:19:20 GMT